Add citation

README.md

@@ -8,7 +8,7 @@
 * The clusters were segmented by seed based distance transformed watershed:
 	- Euclidean Distance Map was calculated for each RoI
 	- Seed points found as maxima on the distance distance map.
-	- Marker controlled watershed was performed using the seed points and the distance
+	- Marker Controlled Watershed [1] was performed using the seed points and the inverted distance transform.
 	
 ![](media/cluster_seg.png) 
 
@@ -16,11 +16,12 @@ This create a list of RoIs for each cell. This allowed each cell to assigned to
 
 ### Detecting of strings
 
-To identify string structure in the GFP channel, for each cell RoI the following was performed:
+To identify string structure in the GFP channel, for each cell RoI and each slice in the Z-stack the following was performed:
 
 - A White Top Hat transformation (MorhphoLibJ [1]) was performed on the GFP channel to enhance string-like features (disk of radius 5px). 
 - The 16-bit image was converted to 8-bit by setting maximum display range to 4 times the 97th percentile of the intensity distribution.  
 - Ridge Detect Plugin [2] was run to create RoIs for each filament
+- The statistics of each filament (length, width etc) were recorded.
 
 
 ![](media/ridge_detection.png)